Cloud computing startup TensorWave is currently focusing on AI technologies and it is making a significant shift. Surprisingly it is not relying on Nvidia’s GPUs as other companies do. It has decided to go with AMD for its latest infrastructure upgrades.
The company has joined a group of specialist cloud operators who have the expertise in managing powerful GPUs and other AI infrastructure. Some players prefer Nvidia while others like CoreWeave and Lambda are opting for AMD.
TensorWave lately has started deploying systems that is powered by AMD’s Instinct MI300X and it is aiming to offer the chips at a lower cost compared to that of Nvidia. Its co-founder Jeff Tatarchuk praised AMD’s latest accelerators due to their availability as well as performance advantages over Nvidia’s offerings.
The company has plans to deploy 20,000 MI300X accelerators across two facilities by the end of this year. It simultaneously has plans to deploy additional liquid-cooled systems in 2025.
AMD’s MI300X comes with impressive specifications and claimed to outperform Nvidia’s highly sought-after H100. It is equipped with advanced design and larger memory capacity. The MI300X is gradually becoming a compelling option for AI workloads.
TensorWave is rapidly installing these systems and leveraging advanced cooling technologies to support dense GPU clusters. It is true that some challenges are being faced like supply chain issues and performance concerns, but the company is optimistic about the potential of AMD’s hardware.
Apart from deploying MI300X nodes, the startup has plans to implement advanced orchestration layers for resource provisioning as well as enhancing the cloud infrastructure’s scalability and performance.
TensorWave is learned to be relying on debt financing for the expansion.
